Asynchronous Task Offloading in Mobile Edge Computing with Uncertain Computation Burden over Multiple Channels

In mobile edge computing (MEC), one of the key issue is to optimize the offloading policy and the allocation of communication and computation resources among multiple mobile users (MUs). For different MUs, deadline of their computation tasks may be heterogeneous. It becomes more challenging as the computation burden of each computation task turns to be a random variable, which may even conform to uncertain probabilistic distribution. To address these issues, this work studies the offloading of asynchronous computation task and resource allocation in a MEC system supporting multiple MUs. Only with the mean and variance about uncertain computation burden, an optimization problem to minimize the weighted sum of energy consumption of multiple MUs is formulated, which is non-deterministic and non-convex, and is hard to solve. To overcome this challenge, we transform it into a deterministic problem, but is still non-convex. In order to solve the non-convex deterministic optimization problem, we decompose the problem into two levels. A heuristic algorithm is proposed for the upper-level to solve an ordering problem and a combination of alternative descend method, successive convex approximation (SCA), and Karush-Kuhn-Tucker (KKT) condition investigating are utilized for the lower-level problem.